JetBrains Air: Building a System of Products for Agentic Software Development

JetBrains Air represents a significant expansion in the company's software development offerings, moving beyond individual developer workbenches to encompass a wider system for agentic development. The era of containing the entire software development process within one window is ending. JetBrains aims to provide developers with an environment where they can work effectively with agents while maintaining control, visibility, and cost management.

At the heart of JetBrains Air is Junie, an AI coding agent designed for professional software development. It will be supported across all Air surfaces, allowing developers to direct agents and verify their output using JetBrains' code intelligence. Air in JetBrains IDEs gives developers the ability to work with agents and review the resulting code.

JetBrains Air also includes Air Teams, which transforms individual agent activity into coordinated team workflows. Air Governance provides visibility, governability, and accountability for agent activity across the organization, ensuring that the work is traceable and overseen. The Agent Client Protocol (ACP) standardizes the connection between an IDE and an agent's full harness, allowing developers to discover and run compatible agents while working within JetBrains IDEs.

Air Governance extends visibility and cost governance across providers and the different tools used for agentic work. This means developers can choose the agent, model, or service best suited for the task without losing context, visibility, or control over the organization. Together, the Air products allow for seamless movement of work between developers, agents, tools, and environments, without losing the essential context and controls.

While developers have been adopting agents faster than organizations can build the necessary infrastructure, the costs associated with agentic development have not kept pace. The hard part lies in verifying the output produced by agents, as code that appears plausible may still contain hidden assumptions or architectural inconsistencies that only surface when it's too expensive to address.

As agents take on more execution responsibilities, the bottleneck shifts from producing change to understanding, verifying, and owning the results. Code becomes cheaper to generate but more expensive to verify. The coordination, audit, and explanation of agent activity become harder, even though the work can still be delegated. Accountability remains with the people and organizations responsible for shipping the software, even as AI improves the quality of the output.

JetBrains Air's multi-vendor support is a foundational design principle, recognizing that the agentic development market is unlikely to consolidate soon. Different models and agents will excel at various tasks, and teams within the same company may make different choices based on their specific needs. Standardizing on a single solution would not serve the diverse requirements of developers and organizations.